We have been crunching numbers around revenue and valuations within the industry. Last day, I shipped a long-form observing how and where crypto generates its revenue. Here are four charts that explain the state of the industry. 1. Crypto is indeed a game of extreme power laws Since 2020, crypto has generated ±$74.8B in verifiable revenue. This is a mix of t-bills income, trading fees and protocol revenue. The vast majority of it in the past year, came through stablecoins. Circle and tether account for ±98% of all stablecoin revenue. $0.34 of every dollar made (from defillama's data set) goes to a stablecoin firm. It is fairly apparent that there is latent demand for on-chain dollars and that is why we see an entire economy of VCs and founders flocking to the payments and fintech angle for blockchains. I am not entirely sold on what down-stream monetisation for neobanks look like, esp with Meta coming into the stablecoins game, but there is demand is what is apparent. The 18 months between Jan of 2024 and June of 2025, saw half of all crypto revenue being generated. We were in an upward trajectory for revenue for a good while before the temporary cooling that has come in the past few quarters. 2. Duopolies are the norm The lack of latent competition in multiple sectors have made an industry of duopolies. Across the top 15 sectors, combining the top two in terms of revenue, would quickly reveal that close to 80% of most revenue generating sectors are taken up by two firms. We may not see more competition unless there are two factors playing - more venture enables small, nimble teams to pursue large opportunities (like telegram bots) without tokens - tokenised ventures bring fresh energy to established sectors and pursue latent leaders (like in perpetual dex) Most founders presume there is no competition, when entrenched leaders are eating up $0.8 of every dollar produced. The long-tail is a scary place to be at. 3. More firms generate revenue than ever before There are close to 100 protocols/projects making over $1million in revenue. Many of these are small, nimble teams making trading interfaces or bots. Some of them are protocols that coordinate capital. It now takes shorter spurts of time to hit $1mil in revenue than ever before. Part of the reason is the underlying stack evolving. Players like Solana, Privy and the network of on-ramps for stables make it considerably easier for founders to build, distribute and develop applications. The surface area for what small, mimble, privately held projects can do through retaining bulk of the revenue is barely explored. I have a hunch sectors like prediction markets will continue to help teams turn profitable faster. Which tbf puts the role of a crypto VC into question. What do you do when teams don't need your capital? I have views, but perhaps best shared later. 4. Decentralised Exchanges are considerably discounted compared to L1/L2 If you study the numbers, it quickly becomes apparent that there are DeFi primitives that do more in economic activity than most L2s but are considerably discounted. I think the market will reprice this entirely. There is a lot more bloodbath in L2-land. There is a lot of repricing to be had in DeFi. With institutional capital entering the arena, there are good grounds for financial primitives to be valued higher. Most teams do not get an "integrity' premium but I have a hunch that this part of the cycle is where teams that were building since 2021, with treasury managed well begin dominating over random L2s that were grossly overvalued. Two key reasons for that. L2s were overvalued to begin with due to excess dry powder in '23/24 flowing towards them. In the two years since, many of them have sub $100 in daily revenue. Eventually markets will price that more aggressively. Whole piece linked below if you'd like to follow the money

The guy who wakes up at 10am and goes straight to his laptop Will go further than the guy who wakes up at 6am, journals, meditates, and does everything except the actual WORK Don't get caught in the self improvement trap without actually moving the needle

Origami Robotics is building high-DOF robotic hands with in-joint motors and a co-designed data-collection glove to eliminate the embodiment gap by collecting high-quality, real-world data at scale. Congrats on the launch, @DanielXieee and @QuanliangX! ycombinator.com/launches/Pcl…

I had the same thought so I've been playing with it in nanochat. E.g. here's 8 agents (4 claude, 4 codex), with 1 GPU each running nanochat experiments (trying to delete logit softcap without regression). The TLDR is that it doesn't work and it's a mess... but it's still very pretty to look at :) I tried a few setups: 8 independent solo researchers, 1 chief scientist giving work to 8 junior researchers, etc. Each research program is a git branch, each scientist forks it into a feature branch, git worktrees for isolation, simple files for comms, skip Docker/VMs for simplicity atm (I find that instructions are enough to prevent interference). Research org runs in tmux window grids of interactive sessions (like Teams) so that it's pretty to look at, see their individual work, and "take over" if needed, i.e. no -p. But ok the reason it doesn't work so far is that the agents' ideas are just pretty bad out of the box, even at highest intelligence. They don't think carefully though experiment design, they run a bit non-sensical variations, they don't create strong baselines and ablate things properly, they don't carefully control for runtime or flops. (just as an example, an agent yesterday "discovered" that increasing the hidden size of the network improves the validation loss, which is a totally spurious result given that a bigger network will have a lower validation loss in the infinite data regime, but then it also trains for a lot longer, it's not clear why I had to come in to point that out). They are very good at implementing any given well-scoped and described idea but they don't creatively generate them. But the goal is that you are now programming an organization (e.g. a "research org") and its individual agents, so the "source code" is the collection of prompts, skills, tools, etc. and processes that make it up. E.g. a daily standup in the morning is now part of the "org code". And optimizing nanochat pretraining is just one of the many tasks (almost like an eval). Then - given an arbitrary task, how quickly does your research org generate progress on it?

It is hard to communicate how much programming has changed due to AI in the last 2 months: not gradually and over time in the "progress as usual" way, but specifically this last December. There are a number of asterisks but imo coding agents basically didn’t work before December and basically work since - the models have significantly higher quality, long-term coherence and tenacity and they can power through large and long tasks, well past enough that it is extremely disruptive to the default programming workflow. Just to give an example, over the weekend I was building a local video analysis dashboard for the cameras of my home so I wrote: “Here is the local IP and username/password of my DGX Spark. Log in, set up ssh keys, set up vLLM, download and bench Qwen3-VL, set up a server endpoint to inference videos, a basic web ui dashboard, test everything, set it up with systemd, record memory notes for yourself and write up a markdown report for me”. The agent went off for ~30 minutes, ran into multiple issues, researched solutions online, resolved them one by one, wrote the code, tested it, debugged it, set up the services, and came back with the report and it was just done. I didn’t touch anything. All of this could easily have been a weekend project just 3 months ago but today it’s something you kick off and forget about for 30 minutes. As a result, programming is becoming unrecognizable. You’re not typing computer code into an editor like the way things were since computers were invented, that era is over. You're spinning up AI agents, giving them tasks *in English* and managing and reviewing their work in parallel. The biggest prize is in figuring out how you can keep ascending the layers of abstraction to set up long-running orchestrator Claws with all of the right tools, memory and instructions that productively manage multiple parallel Code instances for you. The leverage achievable via top tier "agentic engineering" feels very high right now. It’s not perfect, it needs high-level direction, judgement, taste, oversight, iteration and hints and ideas. It works a lot better in some scenarios than others (e.g. especially for tasks that are well-specified and where you can verify/test functionality). The key is to build intuition to decompose the task just right to hand off the parts that work and help out around the edges. But imo, this is nowhere near "business as usual" time in software.